根据土壤盐渍化空间异质性对南疆干旱区绿洲农田进行精准管理分区,对农业种植结构调整和精细化管理具有重要意义。以典型干旱区绿洲农田为研究对象,以电磁感应数据、地形数据和卫星遥感数据为数据源,通过地统计学方法分析研究区土壤盐渍化的空间异质性,利用相关分析筛选出不同时期的植被指数和盐分指数。以农田表层表观电导率（EC_h0.375）为主变量,深层表观电导率（EC_h0.75、EC_v0.75、EC_v1.5）、植被指数（RVI、GRVI、EVI）、土壤盐分指数（NDSI、S5、SI-T）为辅助变量,采用面向对象多尺度分割算法对研究区进行分区,并用分区内平均变异系数（C_V,Coefficient of Variation）和莫兰指数（Moran’s I）对分区结果进行评价。结果表明：（1） 研究区各土层表观电导率都存在明显空间异质性,且各辅助变量与主变量均存在极显著相关。（2） 各分区的平均变异系数较整个研究区的变异系数下降了60%,且基于多源数据分区的区间异质性均高于单一数据分区的区间异质性。（3） 从农田耕作角度、分割效果及分区评价原则出发,综合利用表层和深层土壤盐渍化信息的管理分区效果最佳,该分区结果既符合当地农田管理又符合机械化作业要求。研究结果可为南疆干旱区绿洲农田的精准管理分区提供一定的技术和方法借鉴。

According to the spatial heterogeneity of soil salinization, the precise management precise zoning of oasis farmland in the arid area of Southern Xinjiang is of great significance for the adjustment of areas, based on spatial heterogeneity in soil salinization, is important for determining agricultural planting structure structures and fine-scale management. This study selected the topical oasis farmland in arid zone as the study object and used electromagnetic induction data, topographic date, and satellite remote sensing data as for oasis farmland in the data sources. We analyzed arid zone of Southern Xinjiang to analyze the spatial heterogeneity of soil salinization by geostatistical. Geostatistical methods and screened the were used to evaluate vegetation index and salt index in indices for different periods with correlation analysis. The surface apparent electrical conductivity (EC_h0.375) of the farmland was used as the main variable, and the deep apparent electrical conductivity (EC_h0.75, EC_V0.75, EC_V1.5), vegetation index (RVI, GRVI, EVI), and soil salinity index (NDSI, S5, SI-T) were used as the auxiliary variables. The study area was partitioned by using an object-oriented multi-scale segmentation algorithm, and the zoning results were evaluated by the mean Coefficient of variation (C_V) and Moran’s I. The results showed that there was obvious spatial heterogeneity in the apparent electrical conductivity of each soil layers in the study area, and the auxiliary variables were significantly correlated with the main variable. The average coefficient of variation of each partition was reduced by 60% compared with that of the whole study area, and the. The interval heterogeneity based on multi-source data zoning was higher than that based on single-source data zoning. From the perspective of farmland cultivation, segmentation effects, and zoning evaluation principles, the, management zoning effect that integrated the information of surface and deep soil salinization was the best. And the zoning results were not only consistent with both local farmland management but also met the mechanized operation requirements. The findings can provide certain a technical and methodological reference for precise management zoning of oasis farmland in arid areas of Southern Xinjiang.